EXAM 2 Flashcards
1
Q
RCT
A
Pro’s
- Control for all 3 biases
Con’s
- Feasible
- Generalisability due to high selectivity
- Short follow-up cannot subject people for too long
- Very control impacting the normal evolution of a disease as seen in clinical practice
2
Q
When does selection bias occur?
A
During BOTH obtaining subjects & allocating then to the treatment or control arm.
3
Q
How is selection bias minimised?
A
- Randomisation
- Should generate groups with similar baseline characteristics
- Block randomisation OR 1:1 randomisation - Allocation concealment (blinding)
- Allocation bias
4
Q
What is performance bias?
A
Occurs during implementation of the intervention when either or the administrator or participant is aware which arm of the experiment they are in.
5
Q
How is performance bias minimised?
A
Double blinding
- Blinding both the administrator and participants
- Protects against giving treatment to healthier participants
- Protects against the hawthorn affect –> participants modify their behaviour because they know they are being observed
- Protects against the placebo affect
6
Q
What is attrition bias?
A
- Occurs during participant follow-up
- Patients who drop out are not accounted for - ADRs, motivation, forgot to record data
- Shorter trials = less bias BUT less long term affects
7
Q
How is attrition bias minimised?
A
Attention-to-treat analysis
8
Q
What is detection bias?
A
Measurement OR ascertainment bias
- Occurs during the evaluation of the outcome
- Data assessors know which is the control and treatment arm
9
Q
How is detection bias minimised
A
Blinding the outcome assessors
10
Q
How is risk for binary outcomes measured?
A
- Relative risk
- Relative risk reduction
- Relative risk increase
- Absolute risk
11
Q
What is the relative risk (RR)?
A
Exposed event rate/control event rate
- Cross-sectional, RCTs and cohort studies
- 1 = same risk in both groups
- > 1 = exposure increases risk of the disease or treatment leads to the beneficial outcome
- <1 = exposure decreases risk of the disease
12
Q
What is the relative risk reduction?
A
The treatment decreases your risk of the outcome by X amount
(1-RR) x 100
- Study of therapy
13
Q
What is the relative risk increase?
A
The exposure increases your risk of the outcome by X amount
(RR-1) x 100
- Study of harm
14
Q
What is the absolute risk?
A
> 0 increased risk of the disease outcome (ARI)
<0 decreased risk of the disease outcome (ARR)
15
Q
What is the number needed to treat (NNT)?
A
Number of people you must treat to prevent the outcome
- SMALL number = more lives saved
1/ARR
ARR = absolute risk <0
16
Q
Number needed to harm (NNH)?
A
Number of people that is exposed an outcome will occur.
- LARGE number = less deaths
1/ARI
ARI = absolute risk >0
17
Q
How is risk for continuous outcomes measured?
A
Mean OR mean difference
18
Q
Statistical vs clinical significance
A
Statistical significance is proven by the p-value and CI describing that drug A is different to drug B
Clinical significance is patient oriented based on pain scores and quality of life measures.
19
Q
Superiority trial
A
RCT that aims to prove a new intervention is SUPERIOR to a comparator.
20
Q
Inferiority trial
A
RCT that aims to prove an intervention is EQUAL or superior to an existing intervention.
21
Q
Interim analysis
A
Third party analyses the data to ensure no danger is occurring to the participant.
If done early it may lead to false positives due to the lack of data at that point in time.
22
Q
Cohort study
A
Exposed vs non-exposed cohorts are followed FORWARD in time to asses if they develop the outcome of interest.
Pro’s
- Generalisable
Con’s
- Time consuming
- Costly
- Susceptible to bias (selection, performance, attrition)
- False causation
23
Q
Prospective cohort study
A
Participants are selected at the time the BEGIN the exposure (e.g., starting the OCP today) and then followed FORWARD in time until they develop the outcome of interest (e.g. breast cancer).
24
Q
Retrospective cohort study
A
Participants are selected based on a PAST and CONTINUING exposure (e.g., smoked for 10 years and not gonna quite) and followed FORWARD in time until the outcome of interest is reached (e.g., lung cancer)
- Good for diseases that take time to develop as you are getting a head start on the exposure
25
What is the selection based on for a cohort study
Selected based on the absence or presence of an EXPOSURE and absence of the outcome
26
What bias is not protected for in a cohort study?
Selection & allocation bias
- It is not possible to randomise patients into groups as they know if they have the exposure or not
Performance bias
- Participants know their exposure thus cannot be blinded against this
Attrition bias
- Can be protected for BUT because of their long duration they are vulnerable to extensive drop outs
27
How is risk measured in a cohort study?
Relative risk
- If all the participants reach the outcome at a similar time
Person-time risk & Hazard ratios
- If participants reach the outcome at different times
28
What are confounders?
Prognostic variables
- Independent risk factors associated with BOTH the exposure and the outcome
- Occur due to the absence of randomisation and thus the presence of selection bias
- Corrected for via analytic correction
29
False causation
Occurs in observational studies (cohort and case-control) due to chance events, bias, and/or confounding variables.
False association between the exposure and outcome
30
Test of causation validity in observational studies
1. Did the exposure precede the outcome?
2. Is there a dose-dependent relationship?
3. De-challenge-rechallenge strategy?
- Did other studies have the same outcome?
- Does the association make biological sense?
31
How are results validated in a cohort study and RCT?
1. Same baseline characteristics between groups (selection bias)
2. Detection methods the same (performance and measurement bias)
3. Follow-up long and sufficient (attrition bias)
32
How is precision assessed in RCTs and cohort studies?
Confidence interval
33
How is the strength of an association measured in RCTs and cohort studies?
R-value
34
Case-control study
Participants WITH the outcome of interest (e.g. lung cancer) are surveyed about their PAST exposures to find an association between exposures and the outcome of interest.
Pro's
- Fast
- Asses rare events (not waiting for them to happen)
Con's
- Susceptible to bias
35
What is selection based on for case-control studies
Participants are selected based on having the OUTCOME or not
36
What biases are specific for case-control studies
```
Recall bias (passive)
- Participants cannot remember if they had the exposure
```
```
Response bias (active)
- False response because they did not want to admit their exposure
```
37
What bias is not applicable to case-control studies?
Attrition
| - Participants already have the outcome
38
How is risk measured in case-control studies?
```
Odds ration (OR)
- The odds of developing the disease is (OR) times greater is exposed to X
>1 = harmful exposure
<1 = protective exposure
1 = exposure unlikely to cause harm
```
39
Patient expected event rate (PEER)
Expected outcome rate in unexposed individuals based from general population control event rates (e.g., 10% of Australian men develop prostate cancer), other studies or extrapolated from data.
40
How are results validated in case-control studies?
1. Cases were well defined and selected
2. Information about the exposure was similarly obtained
3. All confounders were accounted for
41
Inception study
Longitudinal study
| - Cohort study whereby participants are selected near the onset of their disease
42
What biases are specific for prognostic studies?
Sampling bias
- Type of selection bias
- Minimised by having a detailed description of the cohort and recruiting them at similar time in their disease
Survival bias
- Occurs when the cohort was selected too late in the disease progression so that those with poor prognostic factors died prior to the study and thus were not included
- Can occur when studied are conducted in the community rather than the hospital where the really sick people are
Non-response bias
- Those who are the most sick may not respond during follow-up
- Protected for by sensitivity analysis whereby the worst and best outcomes are calculated and a range obtained
43
Systematic review
QUALITATIVE analysis of multiple studies.
Pro's
- Up-to-date summary of current knowledge
- Minimal bias
- Identify gaps in knowledge
- Increased power --> combine sample sizes
- Increased precision --> multiple CI
- Rare disease and novel interventions due to combined sample sizes
- Generalisable
Con's
- No gold standard for assessing studies
- Selective reporting of the outcomes
44
Publication bias
Studied with negative results are not published OR published later than positive findings.
- Occurs during article selection thus a type of selection bias AND is due to altercations in publishing so a type of reporting bias
45
Time lag bias
Positive results are published more rapidly
46
Language bias
Positive results are published in English
47
Multiple publication bias
Positive results are published more than once
48
Citation bias
Positive results are cited by others
49
Funnel plots
- Examine the possibility of publication bias
- Effect of the treatment is plotted against the sample size
- No bias = symmetrical funnel
- Apex represents small sample sizes with least precision
- Base represents large sample sizes with acute estimates
50
Narrative review
Type of systematic review which focuses on one specific part of the critical appraisal by a professional in that field. E.g., someone who specialised in bias focuses on bias.
51
Meta-analysis
Statistical synthesis of numerical results from several trails
- May be embedded in a systematic review
52
Forrest plots
- Horizontal line = RR
- Squares = 95% CI
- Square size proportional to the weight of that trial
- Dimond = combined RR and CI
- Vertical line = no treatment effect = RR of 1 = unlikely that clinical heterogenicity exists
53
Chi-squared test
- Tests if statistical heterogenicity is present --> variation between results than cannot be explained by chance
P-value <0.05 = variation present = treatment effect unlikely
Chi stat > degree of freedom = variation present
12 statistic of >75% = variation present
54
Sensitivity analysis for meta-analysis
Acknowledges that some studies are poor quality and some are good, so a trial looking at only the good and only the poor is conducted, respectively, to see the change in results.
55
Risk disclosure
Limb 1/objective
Must disclose material risk = risks that a reasonable person would be likely to attach significance to
Limb 2/subjective
Medical practitioner believes that the particular patient would attach significance to a risk
56
Can a tissue donor pull out at any time?
YES! Must be recorded and kept for 3 years.
57
Does the removal of tissue samples during coronal and hospital post-mortem exams require consent?
No. The sampling of tissue is inferred in the consent process prior to their death.
58
Can the next of Kin reject the coronal investigation.
They can ask the coroner to reconsider. However, there must always be a balance between the public interest in investigating a death and the families interests.
59
What must occur to enable the hospital to have authorisation to examine a body post-mortem?
The patient prior to their death provided WRITTEN or ORAL consent with TWO witnesses
OR - the next of Kin consents
OR - after reasonable efforts to locate the next of kin AND the medical officer believes the patient would not have objected
60
What is the penalty if discarded tissue is used in research without formal consent?
- Fine
- 6 months imprisonment
- or BOTH
61
What are the 4 requirement to determine a patients capacity?
1. Understand their decisions and its effects
2. Retain the information
3. Apply the information and make a decision
4. Communicate why they made the decision
62
What do you do if you are unsure if a patient has capacity?
Seek specialist assessment
63
What if the patient disagrees with your denial of their capacity?
1. They may seek a second opinion
| 2. Challenge via VCAT
64
What is the hierarchy of decision making?
1. Patient with capacity
2. Advanced care directive
3. Medical treatment decision maker
4. Guardian appointed by VCAT
5. Close and continuing relationship with
- Spouse or domestic partner
- Primary carer
- Adult child
- Parent
- Adult sibling
65
What is the next step if an ACD or MTDM cannot be located within a reasonable time after reasonable efforts to locate them?
The doctor must determine if the procedure is routine in which they can administer OR significant which requires the public advocate (surrogate MTDM)
66
What makes a treatment significant?
1. Bodily intrusion
2. Risk
3. Adverse reaction
4. Distress
67
Medical treatment decision maker
Person appointed by the patient when they had capacity.
| Can be multiple - 1st to be contacted
68
What is a value directive?
Patients values to be considered by the MTDM
69
What is a instructive directive?
Objective decision on specific treatments
| Only valid in the state they are constructed - if interstate they are considered values directive
70
Can the advance care plan be rejected?
Yes. If the doctor has reason to believe they are out dated. However, they must apply to VCAT unless the patient is likely to deteriorate.
71
Under what circumstanced can you provide emergency treatment without consent?
1. To save a life
2. Prevent serious damage
3. Prevent suffering
72
Can the MTDM or ACD refuse palliative care?
No. However, they can specify to receive less pain relief as to be more lucid.
73
What treatment can be administered in palliative care?
Pain relief
| Treatment to relief suffering and discomfort
74
What is a support person
Aid a patient make a decision for which they are allowed access to the medical records.
75
Who can apply to VCAT to challenge the MTDM?
- Doctor
- MTDM
- Support person
- Public advocate
- Close and continuing family member
76
What are grounds to challenge the MTDM via VCAT?
1. MTDM is not actually the MTDM
2. MTDM failed to consider the values directive
3. MTDM is not being consistent with the patients preferences
77
What is the definition of suicide?
Patient deliberately kills them self OR patient caused their own death (e.g., accidental OD)
78
Can a patient with capacity refuse treatment?
Yes. If this request is ignored it is ground for battery!
79
How is not for resuscitation and treatment withdrawal decided on?
ACD or medical decision
80
What is the difference between passive and voluntary assisted dying?
Passive is when medical treatment is withdrawn at the patients requests whereas active is when the patients request for a medical intervention to end their life.
81
What are the 5 eligibility criteria for voluntary assisted dying?
1. 18+
2. Australian citizen OR permanent resident
3. Decision making capacity
4. Diagnosed with an incurable disease with <6 month prognosis
5. Experiencing intolerable and unmanageable suffering
Patient must REQUEST it
Cannot be ACD
82
What is the legal definition of brain death?
Irreversible cessation of all brain functions
83
What is legal definition circulatory death?
Irreversible cessation of circulation
- The heart is likely to cease with withdrawal of life-sustaining treatment
- Removal occurs when the heart stops
84
What is involved in the assessment of organ harvesting in someone about to undergo circulatory death?
- Assessment occurs while the heart is still beating on life-sustaining support
1. Neurological assessment
2. TWO doctors not involved in the donation process with >5yrs experience
3. Determination of brain death
85
How is consent obtain for whole organ donation?
WRITTEN or ORAL with TWO witnesses during terminal illness
86
What if whole organ donation is unknown?
1. Ask and gain consent from next of kin
| 2. Designated officer OR doctor has no reason to believe the donor would object
87
What criteria is required for gaining consent for non-regenerative tissue?
1. 18+
2. WRITTEN consent
3. Specify tissue
4. Consent certificate obtain by performing surgeon authorised by a doctor ensuring the donor has voluntary informed consent
24 cooling off prior to harvesting
88
What if the donor for non-regenerative tissue does not have capacity?
Guardian can provide consent or VCAT
89
What criteria is required for regenerative tissue consent?
1. 18+
2. WRITTEN consent (oral okay for blood)
3. consent certificate by the performing doctor and authorised by a doctor confirming informed voluntary consent
90
Can a minor donate non-regenerative tissue?
NO! It is strictly prohibited!
91
How is consent gained for regenerative tissue in a minor?
1. Parent provided WRITTEN consent
2. Only donate for a patient or sibling
3. Dr must certify that the consent was obtained in their presence, parent and minor informed and the minor understood and agreed
92
What if the minor donating regenerative tissue is too young to comprehend?
The parent can consent and the donation can occur if the recipient parent or sibling will most likely die.
93
Can I minor give blood?
Parent can provide consent for medical, scientific or therapeutic reasons if:
1. Dr confirms it will not harm the child
2. Child agrees to removal
94
What are the three components of EBP
1. Clinical expertise
2. Best evidence
3. Patient preferences and values
95
What are the 4 steps in EBCP?
1. Ask
2. Acquire
3. Appraise
4. Apply
96
What is the P-value?
The chance of getting an observed effect if the null hypothesis was true.
<0.05 = significant = there is a 5% chance the result is wrong
Cannot be used in isolation to inform clinical judgement has it does not provide a measure of the size of the effect.
97
How is categorical data graphically represented?
Bar chart or pie chart
98
Ordinal data
Categorical data that is ranked.
e. g., mild, moderate or severe.
- A mean can be generated
99
Nominal data
Categorical data that is classified into groups.
E.g., ABO typing, gender
- No mean can be generated
100
How is numerical data graphically represented?
Histogram or boxplot and is described using the means, median and standard deviation.
101
Discrete variable
Whole numbers
| E.g., Road toll
102
Continuous variables
Fractions and decimals
| E.g., height
103
Pie charts
Represents BOTH nominal and ordinal categorical data
| - Summarises large data sets
104
Bar graph
Represents BOTH nominal and ordinal categorical
- Displays more categories than pie charts
- Shows distribution of data
105
Histogram
Represents only CONTINUOUS numerical data
- Bars are joined to show data continuity
Skew
- Right skew --> tail to the right = positive skew --> mean > median > mode
- Left skew --> tail the left = negative skew --> mode > median > mean
- Median reported
106
Box plot
Represents BOTH continuous and discrete numerical data
- Identify outliers
- Min, Q1, median, Q3, Max
107
Scatter plot
Described the relationship between two variables. The X-axis can be either continuous or discrete but the Y-axis MUST be continuous
108
What is the confidence interval?
- A range in which the true population mean is likely to lie
- Informs on the precision of the study
- Large range = small sample size = poor power
- Small range = large sample size = good power
- Dichotomous --> CI includes 1 = insignificant
- Continuous data --> CI includes 0 = insignificant
109
Normal distribution
- Mean = median = mode
- Symmetrical
- Continuous (numerical) variables only
- Population mean = central value
- SD = spread
68 of the observation fall within 1SD of the mean for the true population
95
99.7
110
Non-parametric tests
- Non-continuous data - categorical
- Not normally distributed i.e., skewed data
- Median and inter quartile range (not mean and SD)
- Not impacted by outliers
- Chi squared
- Fisher's extract
- McNemar (paired)
111
Odd ratio (OR)
- Case-controls
1 = no difference
>1 = increased risk in exposed v non-exposed
<1 = decreased risk in exposed v non-exposed
112
What is the mean?
Considered all data so is affected by outliers
| - Used for normal distribution (symmetrical) NOT skewed data (asymmetrical)
113
What is the standard deviation?
Average distance of observations from the mean
- Affected by extreme values
- Symmetrical data
114
Random error
Error due to chance alone
- Increased with small sample size
- Reduced by repeating measurements
Type 1/false positive
- Null hypothesis is rejected when it is true
- Minimised by the P-value
Type 2/false negative
- Null hypothesis is accepted when it is false
115
What is sensitivity and sensitive tests?
True positive - proportion of patients WITH the disease that have a POSITIVE test
Test detects an abnormal state so can safely RULE OUT disease with a negative test but vulnerable to FALSE POSITIVES
116
What is specificity and what is a specific test
True negative - proportion of patients WITHOUT the disease who have a NEGATIVE test.
Test detects specific disease markers so can safely RULE IN the disease but vulnerable to FALSE NEGATIVES
117
What is the positive predictive value (PPV)?
The probability of having the DISEASE with a POSITIVE test
118
What is the negative predictive value
Probability of NOT having the disease with a NEGATIVE test
119
What is the likelihood ratio positive?
True positive rate/false positive rate
Higher = more likely the result is a true positive
>10
120
What is the likelihood ratio negative?
False negative/true negative
Smaller the LR- the more likely the result is a true negative.
<0.1 = better post-test probability
121
What is the likelihood ratio interpretation?
```
1 = test provided no additional information
>10 = positive test --> certain they have the disease
<0.1 = negative test --> certain they do not have the disease
```
122
What is the post-test probability?
Informs if a diagnostic test should be performed or not.
| Considered the pre-test probability (prevalence) and the test information (predictive values and likelihood ratios)
123
Prognosis
- Measures with HR's
- Survival curves - end of the curve = less observations = wide CI = less precise
- Use RCT, case-control or cohort
